Greetings for the New Year!

I've had the impression member of this group are mostly theoreticians. Hope some of you are interested
in solving real life problems?

I'm the proprietor-interm CEO of Wattminder, a startup web services builder to practice the science of electrical
engineering over the web; to diagnose faults in solar power plants.

I had early exposure to Prof. Zadeh's fuzzy logic, and applying AI to problems of fault detection-diagnosis
in electronic systems. Now I believe AI or knowledge science may be adaptable to our problem of ubiquitous, yet hidden
'under-performance' in most solar power plants. Many of these power plants are operated under 'power purchase
agreement', or lease contracts where the install contractor is the builder-owner-operator of solar power generators
across the country and get their return on investment through actual production of KWh's at each site and sold to property owners at committed rates. Thus optimal performance of each site is their imperative and livelihood. There is very little in term of tools for detecting and diagnosing problems hidden in these power plants, as this industry is still in its infancy stage. Currently small (~3GW world-wide) but is growing at over 100% per year. Revenue is by contract,
based on a portion of KWh generated, at roughly $0.2 per KWh. Income or return on investment is totally dependent on actual production of the sites under operational management.

I have observed first hand how these PPA contractors operate. Maximum use is made of live monitoring of actual production status of the plants, with embellishment in trend, limits, and color coded alarming messages or icons, derived from custom software, or spreadsheet macros. By and large, expert human operators make the call as to any system is experiencing problem, or performing to expectation.

We at Wattminder have developed a web infra-structure; with sensor data-acquisition, electronic fault detector, internet gateway, and a knowledgebase, data-warehouse, conditional probability matrix, live measurement buffer, rudimentary pattern recognition algorithms, etc. We hope to find a few partners among you, to build a robost, heuristic system to develop a rigorous 'under-performance' detector and diagnostic-remedy messaging system.

This is a serious industry problem, which is rather hidden, because or the intractability of Sun's energy from the sky, and many local environmental conditions, as well as design and workmanship in installations. It is not well publicized and the practitioners certainly won't talk about it openly. Except our recent interviewee Mr. Bryan
Banke of Solar Power Partners, see--
http://www.renewablee...

Keeping a low profile, our web site--www.wattminder.com currently only deployed a couple of our IPs in providing a free public service--a simple bench-mark calculator that any visitor or registered user may obtain a benchmark on
demand; as an indirect way to ascertain if a particular solar power site is performing to expectation on a current output checking, compared to our benchmark.

Wattminder was awarded by DOE for a phase 1 study, in November; as a result we are planning on a going out and get engaged with matching partner in hardware, and algorithm implementors to pursue commercialization of our IP's.

THANKS for your attention, and, looking forward to hearing from you!
May the Sun always warm your face and your panels!
-Steve Yang, P.E. aka solarMD, PVsleuth scyang-at-wattminder-dot-com